Publications

2016

2015

O’Reilly, Christian, Jonathan Godbout, Julie Carrier, and Jean-Marc Lina. (2015) 2015. “Combining Time-Frequency and Spatial Information for the Detection of Sleep Spindles.”. Frontiers in Human Neuroscience 9: 70. https://doi.org/10.3389/fnhum.2015.00070.

EEG sleep spindles are short (0.5-2.0 s) bursts of activity in the 11-16 Hz band occurring during non-rapid eye movement (NREM) sleep. This sporadic activity is thought to play a role in memory consolidation, brain plasticity, and protection of sleep integrity. Many automatic detectors have been proposed to assist or replace experts for sleep spindle scoring. However, these algorithms usually detect too many events making it difficult to achieve a good tradeoff between sensitivity (Se) and false detection rate (FDr). In this work, we propose a semi-automatic detector comprising a sensitivity phase based on well-established criteria followed by a specificity phase using spatial and spectral criteria. In the sensitivity phase, selected events are those which amplitude in the 10-16 Hz band and spectral ratio characteristics both reject a null hypothesis (p < 0.1) stating that the considered event is not a spindle. This null hypothesis is constructed from events occurring during rapid eye movement (REM) sleep epochs. In the specificity phase, a hierarchical clustering of the selected candidates is done based on events' frequency and spatial position along the anterior-posterior axis. Only events from the classes grouping most (at least 80%) spindles scored by an expert are kept. We obtain Se = 93.2% and FDr = 93.0% in the first phase and Se = 85.4% and FDr = 86.2% in the second phase. For these two phases, Matthew's correlation coefficients are respectively 0.228 and 0.324. Results suggest that spindles are defined by specific spatio-spectral properties and that automatic detection methods can be improved by considering these features.

O’Reilly, Christian, and Tore Nielsen. (2015) 2015. “Automatic Sleep Spindle Detection: Benchmarking With Fine Temporal Resolution Using Open Science Tools.”. Frontiers in Human Neuroscience 9: 353. https://doi.org/10.3389/fnhum.2015.00353.

Sleep spindle properties index cognitive faculties such as memory consolidation and diseases such as major depression. For this reason, scoring sleep spindle properties in polysomnographic recordings has become an important activity in both research and clinical settings. The tediousness of this manual task has motivated efforts for its automation. Although some progress has been made, increasing the temporal accuracy of spindle scoring and improving the performance assessment methodology are two aspects needing more attention. In this paper, four open-access automated spindle detectors with fine temporal resolution are proposed and tested against expert scoring of two proprietary and two open-access databases. Results highlight several findings: (1) that expert scoring and polysomnographic databases are important confounders when comparing the performance of spindle detectors tested using different databases or scorings; (2) because spindles are sparse events, specificity estimates are potentially misleading for assessing automated detector performance; (3) reporting the performance of spindle detectors exclusively with sensitivity and specificity estimates, as is often seen in the literature, is insufficient; including sensitivity, precision and a more comprehensive statistic such as Matthew's correlation coefficient, F1-score, or Cohen's κ is necessary for adequate evaluation; (4) reporting statistics for some reasonable range of decision thresholds provides a much more complete and useful benchmarking; (5) performance differences between tested automated detectors were found to be similar to those between available expert scorings; (6) much more development is needed to effectively compare the performance of spindle detectors developed by different research teams. Finally, this work clarifies a long-standing but only seldomly posed question regarding whether expert scoring truly is a reliable gold standard for sleep spindle assessment.

Durka, Piotr J, Urszula Malinowska, Magdalena Zieleniewska, Christian O’Reilly, Piotr T Różański, and Jarosław Żygierewicz. (2015) 2015. “Spindles in Svarog: Framework and Software for Parametrization of EEG Transients.”. Frontiers in Human Neuroscience 9: 258. https://doi.org/10.3389/fnhum.2015.00258.

We present a complete framework for time-frequency parametrization of EEG transients, based upon matching pursuit (MP) decomposition, applied to the detection of sleep spindles. Ranges of spindles duration (>0.5 s) and frequency (11-16 Hz) are taken directly from their standard definitions. Minimal amplitude is computed from the distribution of the root mean square (RMS) amplitude of the signal within the frequency band of sleep spindles. Detection algorithm depends on the choice of just one free parameter, which is a percentile of this distribution. Performance of detection is assessed on the first cohort/second subset of the Montreal Archive of Sleep Studies (MASS-C1/SS2). Cross-validation performed on the 19 available overnight recordings returned the optimal percentile of the RMS distribution close to 97 in most cases, and the following overall performance measures: sensitivity 0.63 ± 0.06, positive predictive value 0.47 ± 0.08, and Matthews coefficient of correlation 0.51 ± 0.04. These concordances are similar to the results achieved on this database by other automatic methods. Proposed detailed parametrization of sleep spindles within a universal framework, encompassing also other EEG transients, opens new possibilities of high resolution investigation of their relations and detailed characteristics. MP decomposition, selection of relevant structures, and simple creation of EEG profiles used previously for assessment of brain activity of patients in disorders of consciousness are implemented in a freely available software package Svarog (Signal Viewer, Analyzer and Recorder On GPL) with user-friendly, mouse-driven interface for review and analysis of EEG. Svarog can be downloaded from http://braintech.pl/svarog.

Duval, Thérésa, Céline Rémi, Réjean Plamondon, Jean Vaillant, and Christian O’Reilly. (2015) 2015. “Combining Sigma-Lognormal Modeling and Classical Features for Analyzing Graphomotor Performances in Kindergarten Children.”. Human Movement Science 43: 183-200. https://doi.org/10.1016/j.humov.2015.04.005.

This paper investigates the advantage of using the kinematic theory of rapid human movements as a complementary approach to those based on classical dynamical features to characterize and analyze kindergarten children's ability to engage in graphomotor activities as a preparation for handwriting learning. This study analyzes nine different movements taken from 48 children evenly distributed among three different school grades corresponding to pupils aged 3, 4, and 5 years. On the one hand, our results show that the ability to perform graphomotor activities depends on kindergarten grades. More importantly, this study shows which performance criteria, from sophisticated neuromotor modeling as well as more classical kinematic parameters, can differentiate children of different school grades. These criteria provide a valuable tool for studying children's graphomotor control learning strategies. On the other hand, from a practical point of view, it is observed that school grades do not clearly reflect pupils' graphomotor performances. This calls for a large-scale investigation, using a more efficient experimental design based on the various observations made throughout this study regarding the choice of the graphic shapes, the number of repetitions and the features to analyze.

Nielsen, T, C O’Reilly, M Carr, G Dumel, I Godin, E Solomonova, J Lara-Carrasco, C Blanchette-Carrière, and T Paquette. (2015) 2015. “Overnight Improvements in Two REM Sleep-Sensitive Tasks Are Associated With Both REM and NREM Sleep Changes, Sleep Spindle Features, and Awakenings for Dream Recall.”. Neurobiology of Learning and Memory 122: 88-97. https://doi.org/10.1016/j.nlm.2014.09.007.

Memory consolidation is associated with sleep physiology but the contribution of specific sleep stages remains controversial. To clarify the contribution of REM sleep, participants were administered two REM sleep-sensitive tasks to determine if associated changes occurred only in REM sleep. Twenty-two participants (7 men) were administered the Corsi Block Tapping and Tower of Hanoi tasks prior to and again after a night of sleep. Task improvers and non-improvers were compared for sleep structure, sleep spindles, and dream recall. Control participants (N = 15) completed the tasks twice during the day without intervening sleep. Overnight Corsi Block improvement was associated with more REM sleep whereas Tower of Hanoi improvement was associated with more N2 sleep. Corsi Block improvement correlated positively with %REM sleep and Tower of Hanoi improvement with %N2 sleep. Post-hoc analyses suggest Tower of Hanoi effects-but not Corsi Block effects-are due to trait differences. Sleep spindle density was associated with Tower of Hanoi improvement whereas spindle amplitude correlated with Corsi Block improvement. Number of REM awakenings for dream reporting (but not dream recall per se) was associated with Corsi Block, but not Tower of Hanoi, improvement but was confounded with REM sleep time. This non-replication of one of 2 REM-sensitive task effects challenges both 'dual-process' and 'sequential' or 'sleep organization' models of sleep-dependent learning and points rather to capacity limitations on REM sleep. Experimental awakenings for sampling dream mentation may not perturb sleep-dependent learning effects; they may even enhance them.

O’Reilly, Christian, Isabelle Godin, Jacques Montplaisir, and Tore Nielsen. (2015) 2015. “REM Sleep Behaviour Disorder Is Associated With Lower Fast and Higher Slow Sleep Spindle Densities.”. Journal of Sleep Research 24 (6): 593-601. https://doi.org/10.1111/jsr.12309.

To investigate differences in sleep spindle properties and scalp topography between patients with rapid eye movement sleep behaviour disorder (RBD) and healthy controls, whole-night polysomnograms of 35 patients diagnosed with RBD and 35 healthy control subjects matched for age and sex were compared. Recordings included a 19-lead 10-20 electroencephalogram montage and standard electromyogram, electrooculogram, electrocardiogram and respiratory leads. Sleep spindles were automatically detected using a standard algorithm, and their characteristics (amplitude, duration, density, frequency and frequency slope) compared between groups. Topological analyses of group-discriminative features were conducted. Sleep spindles occurred at a significantly (e.g. t34 = -4.49; P = 0.00008 for C3) lower density (spindles ∙ min(-1) ) for RBD (mean ± SD: 1.61 ± 0.56 for C3) than for control (2.19 ± 0.61 for C3) participants. However, when distinguishing slow and fast spindles using thresholds individually adapted to the electroencephalogram spectrum of each participant, densities smaller (31-96%) for fast but larger (20-120%) for slow spindles were observed in RBD in all derivations. Maximal differences were in more posterior regions for slow spindles, but over the entire scalp for fast spindles. Results suggest that the density of sleep spindles is altered in patients with RBD and should therefore be investigated as a potential marker of future neurodegeneration in these patients.

2014

O’Reilly, Christian, and Tore Nielsen. (2014) 2014. “Assessing EEG Sleep Spindle Propagation. Part 2: Experimental Characterization.”. Journal of Neuroscience Methods 221: 215-27. https://doi.org/10.1016/j.jneumeth.2013.08.014.

BACKGROUND: This communication is the second of a two-part series that describes and tests a new methodology for assessing the propagation of EEG sleep spindles. Whereas the first part describes the methodology in detail, this part proposes a thorough evaluation of the approach by applying it to a sample of laboratory sleep recordings.

NEW METHOD: The tested methodology is based on the alignment of time-frequency representations of spindle activity across recording channels and is used for assessing sleep spindle propagation over the human scalp using noninvasive EEG.

RESULTS: Spindle propagation displays features that suggest wave displacements of global synaptic potential fields. Propagation patterns that are coherent (as opposed to random), laterally symmetrical, and highly repeatable within and between subjects were observed. Propagation was slower from posterior to anterior and from central to lateral brain regions than in the opposite directions. Propagation speeds varying between 2.3 and 7.0m/s were obtained. A distinct grouping of propagation properties was noted for a small cluster of frontal electrodes. No propagation between distantly separated scalp locations was observed. The values of spindle characteristics such as average frequency, RMS amplitude, frequency slope, and duration, depend largely on propagation direction but are only mildly correlated with propagation delay.

COMPARISON WITH EXISTING METHOD(S): Results obtained are in line with many results published in the literature and offer new measures for describing sleep spindle behavior.

CONCLUSIONS: Propagation properties provide new information about sleep spindle behavior and thus allow more precise automated assessments of spindle-related functions.

O’Reilly, Christian, and Tore Nielsen. (2014) 2014. “Assessing EEG Sleep Spindle Propagation. Part 1: Theory and Proposed Methodology.”. Journal of Neuroscience Methods 221: 202-14. https://doi.org/10.1016/j.jneumeth.2013.08.013.

BACKGROUND: A convergence of studies has revealed sleep spindles to be associated with sleep-related cognitive processing and even with fundamental waking state capacities such as intelligence. However, some spindle characteristics, such as propagation direction and delay, may play a decisive role but are only infrequently investigated because of technical complexities.

NEW METHOD: A new methodology for assessing sleep spindle propagation over the human scalp using noninvasive electroencephalography (EEG) is described. This approach is based on the alignment of time-frequency representations of spindle activity across recording channels.

RESULTS: This first of a two-part series concentrates on framing theoretical considerations related to EEG spindle propagation and on detailing the methodology. A short example application is provided that illustrates the repeatability of results obtained with the new propagation measure in a sample of 32 night recordings. A more comprehensive experimental investigation is presented in part two of the series.

COMPARISON WITH EXISTING METHOD(S): Compared to existing methods, this approach is particularly well adapted for studying the propagation of sleep spindles because it estimates time delays rather than phase synchrony and it computes propagation properties for every individual spindle with windows adjusted to the specific spindle duration.

CONCLUSIONS: The proposed methodology is effective in tracking the propagation of spindles across the scalp and may thus help in elucidating the temporal aspects of sleep spindle dynamics, as well as other transient EEG and MEG events. A software implementation (the Spyndle Python package) is provided as open source software.

O’Reilly, Christian, Réjean Plamondon, and Louise-Hélène Lebrun. (2014) 2014. “Linking Brain Stroke Risk Factors to Human Movement Features for the Development of Preventive Tools.”. Frontiers in Aging Neuroscience 6: 150. https://doi.org/10.3389/fnagi.2014.00150.

This paper uses human movement analyses to assess the susceptibility of brain stroke, one of the most important causes of disability in elders. To that end, a computerized battery of nine neuromuscular tests has been designed and evaluated with a sample of 120 subjects with or without stoke risk factors. The kinematics of the movements produced was analyzed using a computational neuromuscular model and predictive characteristics were extracted. Logistic regression and linear discriminant analysis with leave-one-out cross-validation was used to infer the probability of presence of brain stroke risk factors. The clinical potential value of movement information for stroke prevention was assessed by computing area under the receiver operating characteristic curve (AUC) for the diagnostic of risk factors based on motion analysis. AUC mostly varying between 0.6 and 0.9 were obtained, depending on the neuromuscular test and the risk factor investigated (obesity, diabetes, hypertension, hypercholesterolemia, cigarette smoking, and cardiac disease). Our results support the feasibility of the proposed methodology and its potential application for the development of brain stroke prevention tools. Although further research is needed to improve this methodology and its outcome, results are promising and the proposed approach should be of great interest for many experimenters open to novel approaches in preventive medicine and in gerontology. It should also be valuable for engineers, psychologists, and researchers using human movements for the development of diagnostic and neuromuscular assessment tools.